Upload folder using huggingface_hub

.gitattributes

@@ -35,3 +35,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 checkpoints-v5/checkpoint-12288/eval_state.json filter=lfs diff=lfs merge=lfs -text
 checkpoints-v5.1/checkpoint-12288/eval_state.json filter=lfs diff=lfs merge=lfs -text
+checkpoints-v5.2/checkpoint-11264/eval_state.json filter=lfs diff=lfs merge=lfs -text

checkpoints-v5.2/checkpoint-11264/ema.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f2242e440eec3def700ef54a81144370c3db53b96738cabe892f4077e78668f
+size 54599376

checkpoints-v5.2/checkpoint-11264/eval_state.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36f1f4672fd95a45825d62eb47ac7b08bb6ba3249dc46758ac1ffb574e8f9660
+size 44036748

checkpoints-v5.2/checkpoint-11264/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0c28c9d8d56b0f256f779d5c1c3c2d09e5e41b9462bec886e4b8c81c320930a
+size 54599408

checkpoints-v5.2/checkpoint-11264/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:377971d6247a2ad31cdea9a913ab5da9a55b0b5db9068480ab264c26af741097
+size 76550347

checkpoints-v5.2/checkpoint-11264/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45b1ca7f396cc5599f3f64fbc801b07bc6329480ea979bb7f593fcbee2c79ab9
+size 14645

checkpoints-v5.2/checkpoint-11264/scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c07c33eb3e2182286288b9d2aa659daed2b68835c0afb29f8d6c7aee7927a822
+size 1383

checkpoints-v5.2/checkpoint-11264/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7672f8041e5b7bc5a00bd3cd6bedae5ac365dd2bdcbcd4bbaa88eefac435f928
+size 1465

checkpoints-v5.2/checkpoint-11264/trainer_state.json

@@ -0,0 +1,452 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5202531060920974,
+  "eval_steps": 1024,
+  "global_step": 11264,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.047295736917463395,
+      "grad_norm": 3.1038711071014404,
+      "learning_rate": 0.00029970703124999995,
+      "loss": 7.864967346191406,
+      "step": 1024
+    },
+    {
+      "epoch": 0.047295736917463395,
+      "eval_bleu": 0.3418101654679444,
+      "eval_ce_clean_loss": 0.2192277379299952,
+      "eval_ce_pred_loss": 3.7604843192993234,
+      "eval_flow_cos_loss": 0.38796317584166246,
+      "eval_flow_mse_loss": 0.8011491894721985,
+      "eval_loss": 2.99760987987257,
+      "flow/cos_sim": 0.6120368439585107,
+      "flow/improvement_ratio": 0.9814901100173933,
+      "flow/mag_ratio_mean": 0.48378803026458445,
+      "flow/mag_ratio_std": 0.15437949193667058,
+      "step": 1024
+    },
+    {
+      "epoch": 0.047295736917463395,
+      "eval_bleu": 0.3418101654679444,
+      "eval_ce_clean_loss": 0.2192277379299952,
+      "eval_ce_pred_loss": 3.7604843192993234,
+      "eval_flow_cos_loss": 0.38796317584166246,
+      "eval_flow_mse_loss": 0.8011491894721985,
+      "eval_loss": 2.99760987987257,
+      "eval_runtime": 197.0177,
+      "eval_samples_per_second": 142.084,
+      "eval_steps_per_second": 2.223,
+      "flow/cos_sim": 0.6120368439585107,
+      "flow/improvement_ratio": 0.9814901100173933,
+      "flow/mag_ratio_mean": 0.48378803026458445,
+      "flow/mag_ratio_std": 0.15437949193667058,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09459147383492679,
+      "grad_norm": 1.4018265008926392,
+      "learning_rate": 0.00029818297300322,
+      "loss": 2.6447885036468506,
+      "step": 2048
+    },
+    {
+      "epoch": 0.09459147383492679,
+      "eval_bleu": 0.3947018649544751,
+      "eval_ce_clean_loss": 0.03551059079326723,
+      "eval_ce_pred_loss": 3.0670854468323867,
+      "eval_flow_cos_loss": 0.26620999992438105,
+      "eval_flow_mse_loss": 0.7514688725068689,
+      "eval_loss": 2.3870746936972282,
+      "flow/cos_sim": 0.733790022325298,
+      "flow/improvement_ratio": 0.9944704486627013,
+      "flow/mag_ratio_mean": 0.6799977410873866,
+      "flow/mag_ratio_std": 0.11552925763405077,
+      "step": 2048
+    },
+    {
+      "epoch": 0.09459147383492679,
+      "eval_bleu": 0.3947018649544751,
+      "eval_ce_clean_loss": 0.03551059079326723,
+      "eval_ce_pred_loss": 3.0670854468323867,
+      "eval_flow_cos_loss": 0.26620999992438105,
+      "eval_flow_mse_loss": 0.7514688725068689,
+      "eval_loss": 2.3870746936972282,
+      "eval_runtime": 192.2679,
+      "eval_samples_per_second": 145.594,
+      "eval_steps_per_second": 2.278,
+      "flow/cos_sim": 0.733790022325298,
+      "flow/improvement_ratio": 0.9944704486627013,
+      "flow/mag_ratio_mean": 0.6799977410873866,
+      "flow/mag_ratio_std": 0.11552925763405077,
+      "step": 2048
+    },
+    {
+      "epoch": 0.1418872107523902,
+      "grad_norm": 1.5566725730895996,
+      "learning_rate": 0.0002927689070858589,
+      "loss": 2.3424882888793945,
+      "step": 3072
+    },
+    {
+      "epoch": 0.1418872107523902,
+      "eval_bleu": 0.41753887403168577,
+      "eval_ce_clean_loss": 0.017339187425446427,
+      "eval_ce_pred_loss": 2.829169942091589,
+      "eval_flow_cos_loss": 0.14888262367684002,
+      "eval_flow_mse_loss": 0.7409176882271353,
+      "eval_loss": 2.2100625013651913,
+      "flow/cos_sim": 0.8511173925171159,
+      "flow/improvement_ratio": 0.9950010483939898,
+      "flow/mag_ratio_mean": 0.8375713426500695,
+      "flow/mag_ratio_std": 0.10142230264604364,
+      "step": 3072
+    },
+    {
+      "epoch": 0.1418872107523902,
+      "eval_bleu": 0.41753887403168577,
+      "eval_ce_clean_loss": 0.017339187425446427,
+      "eval_ce_pred_loss": 2.829169942091589,
+      "eval_flow_cos_loss": 0.14888262367684002,
+      "eval_flow_mse_loss": 0.7409176882271353,
+      "eval_loss": 2.2100625013651913,
+      "eval_runtime": 194.0365,
+      "eval_samples_per_second": 144.267,
+      "eval_steps_per_second": 2.257,
+      "flow/cos_sim": 0.8511173925171159,
+      "flow/improvement_ratio": 0.9950010483939898,
+      "flow/mag_ratio_mean": 0.8375713426500695,
+      "flow/mag_ratio_std": 0.10142230264604364,
+      "step": 3072
+    },
+    {
+      "epoch": 0.18918294766985358,
+      "grad_norm": 1.6979767084121704,
+      "learning_rate": 0.00028389952193475995,
+      "loss": 2.209465742111206,
+      "step": 4096
+    },
+    {
+      "epoch": 0.18918294766985358,
+      "eval_bleu": 0.4338824863011538,
+      "eval_ce_clean_loss": 0.010503773240491612,
+      "eval_ce_pred_loss": 2.7355331310398503,
+      "eval_flow_cos_loss": 0.10729855863743176,
+      "eval_flow_mse_loss": 0.7243417862071294,
+      "eval_loss": 2.1294367664480864,
+      "flow/cos_sim": 0.8927014652452513,
+      "flow/improvement_ratio": 0.9945996228146227,
+      "flow/mag_ratio_mean": 0.8834205535176682,
+      "flow/mag_ratio_std": 0.09435187315900032,
+      "step": 4096
+    },
+    {
+      "epoch": 0.18918294766985358,
+      "eval_bleu": 0.4338824863011538,
+      "eval_ce_clean_loss": 0.010503773240491612,
+      "eval_ce_pred_loss": 2.7355331310398503,
+      "eval_flow_cos_loss": 0.10729855863743176,
+      "eval_flow_mse_loss": 0.7243417862071294,
+      "eval_loss": 2.1294367664480864,
+      "eval_runtime": 193.5894,
+      "eval_samples_per_second": 144.6,
+      "eval_steps_per_second": 2.263,
+      "flow/cos_sim": 0.8927014652452513,
+      "flow/improvement_ratio": 0.9945996228146227,
+      "flow/mag_ratio_mean": 0.8834205535176682,
+      "flow/mag_ratio_std": 0.09435187315900032,
+      "step": 4096
+    },
+    {
+      "epoch": 0.236478684587317,
+      "grad_norm": 1.5808225870132446,
+      "learning_rate": 0.00027177281107320826,
+      "loss": 2.1267242431640625,
+      "step": 5120
+    },
+    {
+      "epoch": 0.236478684587317,
+      "eval_bleu": 0.4571689702049482,
+      "eval_ce_clean_loss": 0.007218487369381401,
+      "eval_ce_pred_loss": 2.582534832224998,
+      "eval_flow_cos_loss": 0.08806605126759777,
+      "eval_flow_mse_loss": 0.7072322978276641,
+      "eval_loss": 2.0277347072074403,
+      "flow/cos_sim": 0.9119339856110751,
+      "flow/improvement_ratio": 0.9952437794643995,
+      "flow/mag_ratio_mean": 0.9055254207626325,
+      "flow/mag_ratio_std": 0.0895098245205128,
+      "step": 5120
+    },
+    {
+      "epoch": 0.236478684587317,
+      "eval_bleu": 0.4571689702049482,
+      "eval_ce_clean_loss": 0.007218487369381401,
+      "eval_ce_pred_loss": 2.582534832224998,
+      "eval_flow_cos_loss": 0.08806605126759777,
+      "eval_flow_mse_loss": 0.7072322978276641,
+      "eval_loss": 2.0277347072074403,
+      "eval_runtime": 193.0062,
+      "eval_samples_per_second": 145.037,
+      "eval_steps_per_second": 2.269,
+      "flow/cos_sim": 0.9119339856110751,
+      "flow/improvement_ratio": 0.9952437794643995,
+      "flow/mag_ratio_mean": 0.9055254207626325,
+      "flow/mag_ratio_std": 0.0895098245205128,
+      "step": 5120
+    },
+    {
+      "epoch": 0.2837744215047804,
+      "grad_norm": 1.4207789897918701,
+      "learning_rate": 0.00025670620574270894,
+      "loss": 2.07169508934021,
+      "step": 6144
+    },
+    {
+      "epoch": 0.2837744215047804,
+      "eval_bleu": 0.4623355951643849,
+      "eval_ce_clean_loss": 0.005376053338000747,
+      "eval_ce_pred_loss": 2.5099684372884496,
+      "eval_flow_cos_loss": 0.07843345983727881,
+      "eval_flow_mse_loss": 0.7116553213226197,
+      "eval_loss": 1.9916239587683655,
+      "flow/cos_sim": 0.9215665681449245,
+      "flow/improvement_ratio": 0.9947730746443413,
+      "flow/mag_ratio_mean": 0.9184351398792441,
+      "flow/mag_ratio_std": 0.08669416432919567,
+      "step": 6144
+    },
+    {
+      "epoch": 0.2837744215047804,
+      "eval_bleu": 0.4623355951643849,
+      "eval_ce_clean_loss": 0.005376053338000747,
+      "eval_ce_pred_loss": 2.5099684372884496,
+      "eval_flow_cos_loss": 0.07843345983727881,
+      "eval_flow_mse_loss": 0.7116553213226197,
+      "eval_loss": 1.9916239587683655,
+      "eval_runtime": 196.543,
+      "eval_samples_per_second": 142.427,
+      "eval_steps_per_second": 2.229,
+      "flow/cos_sim": 0.9215665681449245,
+      "flow/improvement_ratio": 0.9947730746443413,
+      "flow/mag_ratio_mean": 0.9184351398792441,
+      "flow/mag_ratio_std": 0.08669416432919567,
+      "step": 6144
+    },
+    {
+      "epoch": 0.3310701584222438,
+      "grad_norm": 3.3632595539093018,
+      "learning_rate": 0.0002390360415767374,
+      "loss": 2.014681816101074,
+      "step": 7168
+    },
+    {
+      "epoch": 0.3310701584222438,
+      "eval_bleu": 0.4595571656999387,
+      "eval_ce_clean_loss": 0.004155274415994701,
+      "eval_ce_pred_loss": 2.4867921506977515,
+      "eval_flow_cos_loss": 0.0699176051351986,
+      "eval_flow_mse_loss": 0.6959788431859997,
+      "eval_loss": 1.9610095941312782,
+      "flow/cos_sim": 0.9300824132956327,
+      "flow/improvement_ratio": 0.9952014009702151,
+      "flow/mag_ratio_mean": 0.9340966411921532,
+      "flow/mag_ratio_std": 0.08523170804800508,
+      "step": 7168
+    },
+    {
+      "epoch": 0.3310701584222438,
+      "eval_bleu": 0.4595571656999387,
+      "eval_ce_clean_loss": 0.004155274415994701,
+      "eval_ce_pred_loss": 2.4867921506977515,
+      "eval_flow_cos_loss": 0.0699176051351986,
+      "eval_flow_mse_loss": 0.6959788431859997,
+      "eval_loss": 1.9610095941312782,
+      "eval_runtime": 195.7842,
+      "eval_samples_per_second": 142.979,
+      "eval_steps_per_second": 2.237,
+      "flow/cos_sim": 0.9300824132956327,
+      "flow/improvement_ratio": 0.9952014009702151,
+      "flow/mag_ratio_mean": 0.9340966411921532,
+      "flow/mag_ratio_std": 0.08523170804800508,
+      "step": 7168
+    },
+    {
+      "epoch": 0.37836589533970716,
+      "grad_norm": 1.9611129760742188,
+      "learning_rate": 0.00021922485637411153,
+      "loss": 1.9802610874176025,
+      "step": 8192
+    },
+    {
+      "epoch": 0.37836589533970716,
+      "eval_bleu": 0.4797964191826387,
+      "eval_ce_clean_loss": 0.00341097121557527,
+      "eval_ce_pred_loss": 2.4153971993215553,
+      "eval_flow_cos_loss": 0.06374154817398008,
+      "eval_flow_mse_loss": 0.6859119383439626,
+      "eval_loss": 1.9129569013369139,
+      "flow/cos_sim": 0.936258481245607,
+      "flow/improvement_ratio": 0.9943351938844271,
+      "flow/mag_ratio_mean": 0.9286861550318052,
+      "flow/mag_ratio_std": 0.08223920938086836,
+      "step": 8192
+    },
+    {
+      "epoch": 0.37836589533970716,
+      "eval_bleu": 0.4797964191826387,
+      "eval_ce_clean_loss": 0.00341097121557527,
+      "eval_ce_pred_loss": 2.4153971993215553,
+      "eval_flow_cos_loss": 0.06374154817398008,
+      "eval_flow_mse_loss": 0.6859119383439626,
+      "eval_loss": 1.9129569013369139,
+      "eval_runtime": 193.8032,
+      "eval_samples_per_second": 144.44,
+      "eval_steps_per_second": 2.26,
+      "flow/cos_sim": 0.936258481245607,
+      "flow/improvement_ratio": 0.9943351938844271,
+      "flow/mag_ratio_mean": 0.9286861550318052,
+      "flow/mag_ratio_std": 0.08223920938086836,
+      "step": 8192
+    },
+    {
+      "epoch": 0.4256616322571706,
+      "grad_norm": 1.6621181964874268,
+      "learning_rate": 0.00019771490022228366,
+      "loss": 1.9591320753097534,
+      "step": 9216
+    },
+    {
+      "epoch": 0.4256616322571706,
+      "eval_bleu": 0.4845638244135935,
+      "eval_ce_clean_loss": 0.0029391252050882197,
+      "eval_ce_pred_loss": 2.36602204493736,
+      "eval_flow_cos_loss": 0.061222414167164126,
+      "eval_flow_mse_loss": 0.6859568288064983,
+      "eval_loss": 1.8872125840078207,
+      "flow/cos_sim": 0.9387776077338005,
+      "flow/improvement_ratio": 0.9949078940909747,
+      "flow/mag_ratio_mean": 0.9425890755979982,
+      "flow/mag_ratio_std": 0.08194239209639971,
+      "step": 9216
+    },
+    {
+      "epoch": 0.4256616322571706,
+      "eval_bleu": 0.4845638244135935,
+      "eval_ce_clean_loss": 0.0029391252050882197,
+      "eval_ce_pred_loss": 2.36602204493736,
+      "eval_flow_cos_loss": 0.061222414167164126,
+      "eval_flow_mse_loss": 0.6859568288064983,
+      "eval_loss": 1.8872125840078207,
+      "eval_runtime": 193.8927,
+      "eval_samples_per_second": 144.374,
+      "eval_steps_per_second": 2.259,
+      "flow/cos_sim": 0.9387776077338005,
+      "flow/improvement_ratio": 0.9949078940909747,
+      "flow/mag_ratio_mean": 0.9425890755979982,
+      "flow/mag_ratio_std": 0.08194239209639971,
+      "step": 9216
+    },
+    {
+      "epoch": 0.472957369174634,
+      "grad_norm": 1.4548691511154175,
+      "learning_rate": 0.00017506922209339053,
+      "loss": 1.9363099336624146,
+      "step": 10240
+    },
+    {
+      "epoch": 0.472957369174634,
+      "eval_bleu": 0.48901069646205675,
+      "eval_ce_clean_loss": 0.0027231096204202282,
+      "eval_ce_pred_loss": 2.3451035591565312,
+      "eval_flow_cos_loss": 0.05783578853833077,
+      "eval_flow_mse_loss": 0.6834459889969325,
+      "eval_loss": 1.8731798227519205,
+      "flow/cos_sim": 0.9421642368242621,
+      "flow/improvement_ratio": 0.9953297391601893,
+      "flow/mag_ratio_mean": 0.9415649660101765,
+      "flow/mag_ratio_std": 0.08121724838399452,
+      "step": 10240
+    },
+    {
+      "epoch": 0.472957369174634,
+      "eval_bleu": 0.48901069646205675,
+      "eval_ce_clean_loss": 0.0027231096204202282,
+      "eval_ce_pred_loss": 2.3451035591565312,
+      "eval_flow_cos_loss": 0.05783578853833077,
+      "eval_flow_mse_loss": 0.6834459889969325,
+      "eval_loss": 1.8731798227519205,
+      "eval_runtime": 192.701,
+      "eval_samples_per_second": 145.266,
+      "eval_steps_per_second": 2.273,
+      "flow/cos_sim": 0.9421642368242621,
+      "flow/improvement_ratio": 0.9953297391601893,
+      "flow/mag_ratio_mean": 0.9415649660101765,
+      "flow/mag_ratio_std": 0.08121724838399452,
+      "step": 10240
+    },
+    {
+      "epoch": 0.5202531060920974,
+      "grad_norm": 1.612707495689392,
+      "learning_rate": 0.00015179334717587922,
+      "loss": 1.9111002683639526,
+      "step": 11264
+    },
+    {
+      "epoch": 0.5202531060920974,
+      "eval_bleu": 0.49133817797781537,
+      "eval_ce_clean_loss": 0.00239595649277549,
+      "eval_ce_pred_loss": 2.3280042031584265,
+      "eval_flow_cos_loss": 0.05585474443524124,
+      "eval_flow_mse_loss": 0.6858290074622795,
+      "eval_loss": 1.8661907554761459,
+      "flow/cos_sim": 0.9441452740806423,
+      "flow/improvement_ratio": 0.994876034727924,
+      "flow/mag_ratio_mean": 0.9411461739779607,
+      "flow/mag_ratio_std": 0.08062856834909143,
+      "step": 11264
+    },
+    {
+      "epoch": 0.5202531060920974,
+      "eval_bleu": 0.49133817797781537,
+      "eval_ce_clean_loss": 0.00239595649277549,
+      "eval_ce_pred_loss": 2.3280042031584265,
+      "eval_flow_cos_loss": 0.05585474443524124,
+      "eval_flow_mse_loss": 0.6858290074622795,
+      "eval_loss": 1.8661907554761459,
+      "eval_runtime": 193.6778,
+      "eval_samples_per_second": 144.534,
+      "eval_steps_per_second": 2.261,
+      "flow/cos_sim": 0.9441452740806423,
+      "flow/improvement_ratio": 0.994876034727924,
+      "flow/mag_ratio_mean": 0.9411461739779607,
+      "flow/mag_ratio_std": 0.08062856834909143,
+      "step": 11264
+    }
+  ],
+  "logging_steps": 1024,
+  "max_steps": 21651,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1024,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 64,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoints-v5.2/checkpoint-11264/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e76013f70108cb4ddda27ec8328cf4ef6edec61c75334ae3d6e75eb2e082915f
+size 5137